ics is 510,
not 503, for every book is assigned to the _most specific head that will
contain it_, so that 503 is limited to Dictionaries or Cyclopedias of
Science _in general_. In the same way a General Cyclopedia or Periodical
treats of no one class, and so is assigned to the Class 0. These books
treating of no special class, but general in their character, are divided
into Cyclopedias, Periodicals, etc. No difficulty is found in following
the arithmetical law and omitting the initial zero, so these numbers are
printed 31, 32, etc., instead of 031, 032, etc.

The selection and arrangement of the thousand headings of the
classification cannot be explained in detail for want of space. In all
the work, philosophical theory and accuracy have been made to yield
to practical usefulness. The impossibility of making a satisfactory
classification of all knowledge as preserved in books, has been
appreciated from the first, and nothing of the kind attempted.
Theoretical harmony and exactness has been repeatedly sacrificed to
the practical requirements of the library or to the convenience of the
department in the college. As in every scheme, many minor subjects have
been put under general heads to which they do not strictly belong. In
some cases these headings have been printed in a distinctive type, e. g.,
429 =Anglo-Saxon=, under =ENGLISH PHILOLOGY=. The rule has been to assign
these subjects to the most nearly allied heads, or where it was thought
they would be most useful. The only alternative was to omit them
altogether. If any such omission occurs, it is unintentional and will be
supplied as soon as discovered. Wherever practicable the heads have been
so arranged that each subject is preceded and followed by the most nearly
allied subjects and thus the greatest convenience is secured both in
the catalogues and on the shelves. Theoretically, the division of every
subject into just nine heads is absurd. Practically, it is desirable that
the classification be as minute as possible without the use of additional
figures, and the decimal principle on which our scheme hinges allows nine
divisions as readily as a less number. This principle has proved
wholly satisfactory in practice though it appears to destroy proper
co-ordination in some places. It has seemed best in our library to use
uniformly three figures in the class number. This enables us to classify
certain subjects very minutely, giving, for example, an entire section to